stability.
Khatib, Alfi
Citation
Khatib, A. (2006, October 10). Studies of iso-alpha-acids: analysis,
purification, and stability. Retrieved from https://hdl.handle.net/1887/4860
Version:
Corrected Publisher’s Version
License:
Licence agreement concerning inclusion of doctoral
thesis in the Institutional Repository of the University
of Leiden
Downloaded from:
https://hdl.handle.net/1887/4860
Application of Two Dimensional J-Resolved Nuclear
Magnetic Resonance Spectroscopy to Differentiation of
Beer
Alfi Khatib 1, Erica G. Wilson 1, Hye Kyong Kim 1, Alfons, W. M. Lefeber 2, Cornelis Erkelens 2,Young Hae Choi 1, Robert Verpoorte 1
1
Division of Pharmacognosy, Section of Metabolomics, Institute of Biology, Leiden University, Einsteinweg 55, PO BOX 9502, 2300 RA Leiden, The Netherlands
2
Division of NMR, Leiden Institute of Chemistry, Gorlaeus Laboratories, Einsteinweg 55, P.O. Box 9502, 2300 RA Leiden, The Netherlands
ABSTRACT
A number of ingredients in beer that directly or indirectly affect its quality require an unbiased wide-spectrum analytical method that allows for the determination of a wide array of compounds for its efficient control. 1H nuclear magnetic resonance (NMR) spectroscopy is a method that clearly meets this description as the broad range of compounds in beer is detectable. However, the resulting congestion of signals added to the low resolution of 1H NMR spectra makes the identification of individual components very difficult. Among two-dimensional (2D) NMR techniques that increase the resolution, J-resolved NMR spectra were successfully applied to the analysis of 2-butanol extracts of beer as overlapping signals in 1H NMR spectra were fully resolved by the additional axis of the coupling constant. Principal component analysis based on the projected J-resolved NMR spectra showed a clear separation between all of the six brands of pilsner beer evaluated in this study. The compounds responsible for the differentiation were identified by 2D NMR spectra including correlated spectroscopy and heteronuclear multiple bond correlation spectra together with J-resolved spectra. They were identified as nucleic acid derivatives (adenine, uridine and xanthine), amino acids (tyrosine and proline), organic acid (succinic and lactic acid), alcohols (tyrosol and isopropanol), cholines and carbohydrates. Unfortunately, the presence of iso-α-acids can not be detected by applying this method.
7.1. INTRODUCTION
Beer is a fermented beverage made from malted barley, hop, yeast and water. There are several factors which affect the character of beer, such as ingredients, production processes and storage conditions (Baxter and Hughes, 2001). Even minor differences in those factors can largely influence the taste and quality of beer. Thus, an investigation into the complete chemical composition of beer has become an important issue for evaluating beer quality.
The control of the quality of this type of product is aimed at providing a guarantee on one side of a certain flavour and fragrance which is characteristic of a particular brand and on the present, its stability during the established shelf-time. Furthermore, it must ensure that these parameters are similar if not identical from batch to batch. It is a known fact that up to the moment, no single compound has been found to be responsible for these characteristics, but rather to a very complex association of diverse compounds, the presence or absence of which, and/or their relative content will lead to particular and unique flavours. A very interesting approach to deal with this problem is to develop a method which could provide an accurate analysis of the composition of the whole matrix of the product, or if that were not possible, of at least a part of it which reflects and identifies each compound. The differences and/or similarities would therefore allow for the establishment of a certain pattern related to very particular characteristics of brews or brands.
Stevens et al., 1999), vitamins (Andrés-Lacueva et al., 1998), carboxylic acids (Miwa and Yamamoto, 1996), sulphur compounds (Hill and Smith, 2000) and bitter acids (Hermans-Lokkerbol and Verpoorte, 1994a; Raumschuh et al., 1999), using a DAD or MS detector for the qualitative and quantitative analysis of these components. Capillary electrophoresis has also been used recently for the determination of amino acids, bitter acids, carbohydrates, peptides, phenolic acids, proteins and vitamins in beer (Cortacero-Ramírez et al., 2003). Enzyme linked sorbent analysis has a limited application for protein analysis (Rasmussen, 1998).
Despite this wide array of analytical methods published before, they all have the characteristic of being applicable to the analysis of a certain group of compounds which are present in beer and are known or proposed to be partially responsible for certain characteristics. None of these methods, as such, can be applied for the simultaneous determination of a wide range of compounds and therefore are no suited to be used to obtain a profile of the extremely complex matrix presented by beer. An efficient method should provide a chemical fingerprint i.e., covering a broad range of metabolites. Such a method would be very useful for the quality control of beer and for the evaluation of the impact of any change in the production process. Such a method should be simple and fast – ensuring in this way the absence of any deterioration of sample components – and reproducible in order to be used at any time and any place as a tool to evaluate different samples. These requirements are true for all analytical methods, but especially for a method in which not all the detected compounds are necessarily identified, with a highly reproducible method these compounds might be identified at a later stage. Nuclear magnetic resonance (NMR) spectroscopy is such a techniques that meet these requirements. Taking into account the considerations mentioned above regarding the possible variables that affect the quality, such as the ratio of certain components, the quantitative information provided by NMR is better than that provided by all other methods mentioned as it is the only one in which the intensity of all the signals is directly correlated to the molar concentration. It means that the amount of all compounds can be directly compared, eliminating the need for calibration curves for each individual compound. NMR also has the great advantage of high reproducibility as the spectra are based on the physical properties of a molecule.
cover a broad range of compounds as a rapid and informative quality control tool. However, despite the undoubted attraction of NMR, the complexity of the 1H NMR spectra is a drawback for its use in the analysis of beer. The complexity of the resulting spectrum makes any attempt to identify the components very difficult. In fact, the 1H NMR spectrum of beer shows a predominance of strongly overlapping signals arising from carbohydrates and amino acids (Duarte et al., 2002). This problem encountered with the application of one-dimensional (1D) 1H NMR can be solved in part by two-dimensional (2D) NMR methods. Diverse pulse sequencing techniques make it possible to develop numerous 2D NMR methods. Among the available two-dimensional NMR methods, the use of J-resolved spectroscopy considerably increases the resolution of the signals with short measuring time compared to other 2D NMR methods (Viant, 2003). In the J-resolved spectrum the spin – spin couplings are dispersed along the second axis. These additional variables largely improve the resolution of the NMR spectrum of a mixture.
The purpose of this study is to apply the 2D J-resolved NMR spectroscopic method to evaluate its potential as a method for differentiation of beer. For this, six kinds of pilsner beer were analyzed by NMR spectroscopy combined with principal component analysis.
7.2. MATERIALS AND METHODS
7.2.1. Samples
Six different brands of commercial pilsner beer were obtained from the domestic market in the Netherlands.
7.2.2. Solvents and chemicals
First grade chloroform, ethyl acetate, and 2-butanol were purchased from Merck Biosolve Ltd. (Valkenswaard, the Netherlands). CDCl3 (99.9%), D2O (99.9%), and
(Paris, France). Hexamethyldisiloxane (HMDSO) and trimethylsilane propionic acid sodium salt (TSP) were obtained from Merck (Darmstadt, Germany).
7.2.3. Preparation of samples for NMR
Sample preparation was carried out using two different methods. For direct measurement, beer samples were sonicated for 15 minutes for their degasification. One hundred microliter of D2O containing 0.01% (w/w) TSP were added to 900 µL of
the degassed sample and transferred to a NMR tube.
For liquid–liquid fractionation, 50 mL of a degassed sample was transferred to a 250 mL-flask. Fractionation was carried out by extractions with 50 mL portions of chloroform, ethyl acetate and 2-butanol, vortexing for 1 minute in each case. The organic fractions were obtained by decantation after 1 hour equilibrium, collected separately in a 250 mL-round bottom flask and taken to dryness with a rotary vacuum evaporator at room temperature.
7.2.4. NMR measurements
1H NMR and J-resolved spectra were recorded at 25 °C on a 400 MHz Bruker
AV-400 spectrometer operating at a proton NMR frequency of AV-400.13 MHz. Each 1H NMR spectrum consisted of 128 scans requiring 10 minutes acquisition time with the following parameters: 0.25 Hz/point, pulse width (PW) = 30 (4.0 µs), and relaxation delay (RD) = 5.0 s. A pre-saturation sequence was used to suppress the residual water signal with low power selective irradiation at the water frequency during the recycle delay. FIDs were Fourier transformed with LB = 0.3 Hz and the spectra were zero-filled to 32 K points. The resulting spectra were manually phased and baseline corrected, all using XWIN NMR (Version 3.5, Bruker).
Bruker). Data were exported as the 1D projection (F2 axis) of the 2D J-resolved spectra. 1H–1H-correlated spectroscopy (COSY) spectra were acquired with 1.0 s relaxation delay, 4194 Hz spectral width in both dimensions. The heteronuclear multiple bond correlation (HMBC) spectra were obtained with 1.0 s relaxation delay, 4401 Hz spectral width in F2 and 14500 Hz in F1.
The spectra were referenced to the solvent signal of CDCl3 at δ 7.26, MeOD at δ
3.30, DMSO-d6 at δ 2.49 for the organic solvent extracts and trimethylsilane
propionic acid sodium salt (TSP) at δ 0.00 for aqueous extracts.
7.2.5. Data analysis
The 1H NMR and the J-resolved projection spectra were automatically reduced to ASCII files using AMIX (v. 3.7, Bruker Biospin). Spectral intensities were scaled to total intensity and reduced to integrated regions of equal width (δ 0.04) corresponding to the region of δ 0.40–10.00. The region of δ 4.70–5.10 was excluded from the analysis because of the residual signal of water. Principal component analysis (PCA) were performed with the SIMCA-P software (v. 10.0, Umetrics, Umeå, Sweden) using mean centered scaling method. Dimension of data matrices was 241 × 18.
7.3. RESULTS AND DISCUSSION
7.3.1. Evaluation of different sample preparation methods.
extracts exhibited the highest intensity and best resolution of aromatic signals in the NMR (Fig. 7.1B). However, the signals still showed considerable overlapping. The congestion of NMR signals was finally solved by applying 2D J-resolved NMR spectroscopy.
Fig. 7.1. 400 MHz 1H-NMR spectrum of beer obtained by direct measurement (A) and 2-butanol extract (B) in the range of δ 0.5 – δ 9.0. Expanded spectra cover the δ 5.5 – δ 9.0 range.
A
Projected J-resolved spectra were recorded in order to reduce the complexity of 1H NMR spectra (Fig. 7.2), as all protons ideally appear as a singlet in the J-resolved spectra using the projected spectrum on the axis of chemical shift (just like 13C NMR spectra showing a fully H decoupled spectrum). The resulting enhanced resolution made it possible to directly compare and match the spectra obtained from different samples. The projected J-resolved spectra were applied to discriminate the beer samples through principal component analysis.
Fig. 7.2. 400 MHz 2D J-resolved (A) and its projected (B) spectra of 2-butanol extract of beer in the range of δ 5.5 – δ 9.0. 1; H-2 of adenine, 2; H-5 of adenine, 3; H-3 of uridine, 4; H-2 of uridine, 5; H-2 of xanthine, 6; H-2 and H-6 of tyrosine, 7; H-3 and H-5 of tyrosine, 8; H-2 and H-6 of tyrosol, 9; H-3 and H-5 of tyrosol.
7.3.2. Identification of minor compounds in beer using NMR
The identification of the compounds in the beer samples was performed using 1H NMR, J-resolved, COSY and HMBC spectra. Chemical shifts and coupling constants of the compounds are summarized in Table 7.1. and the chemical structures are shown in Fig. 7.3.
Table 7.1. 1H–NMR chemical shifts and coupling constants of the compounds identified in pilsner beer.
Compound Proton number Chemical shift (ppm) Multiplicity 1 Coupling constant (Hz) Adenine 2 8.32 s - 5 8.19 s - Uridine 3 8.00 d 8.0 2 5.72 d 8.0 1’ 5.90 d Xanthine 2 7.90 s - Tyrosine 2,6 7.12 d 8.5 3,5 6.76 d 8.5 Tyrosol 2,6 7.02 d 8.5 3,5 6.70 d 8.5 7 3.05 t 5.1 Carbohydrates 3.00-4.00 Cholines N-Methyl 3.20, 3.28 s - Succinic acid 2,3 2.56 s - Proline 2 4.03 m - 3 2.43, 2.12 m - 4 2.00 m - Lactic acid 2 4.18 q 7.0 3 1.36 d 7.0 Isopropanol 1,3 1.12 d 8.0
Fig. 7.3. Chemical structures of compounds identified in beer
Fig. 7.4. HMBC spectrum of 2-butanol extract of beer in the aromatic region. 1; H-2/C-8 of adenine, 2; H-2/C-9 of adenine, 3; H-5/C-9 of adenine, 4; H-5/C-10 of adenine, 5; 3/C-1’ of uridine, 6; 3/C-2 of uridine, 7; 3/C-5 of uridine, 8; H-3/C-1 of uridine, 9; H-2/C-3 of uridine, 10; H-1’/C-2’ of uridine, 11; H-1’/C-3 of uridine, 12; H-2/C-8 of xanthine, 13; H-2 and H-6/C-3 and C-5 of tyrosine, 14; H-2 and 6/C-2 and C-6 of tyrosine, 15; 2 and 6/C-4 of tyrosine, 16; 3 and H-5/C-4 of tyrosine; 17; H-2 and H-6/C-3 and C-5 of tyrosol, 18; H-2 and H-6/C-2 and C-6 of tyrosol, 19; H-2 and H-6/C-4 of tyrosol, 20; H-3 and H-5/C-3 and C-5 of tyrosol, 21; H-3 and H-5/C-2 and C-6 of tyrosol, 22; H-3 and H-5/C-4 of tyrosol. In the aliphatic region, cholines, succinic acid, proline, lactic acid and isopropanol are identified in the NMR spectra of 2-butanol extracts of beer. Choline analogues show a singlets at δ 3.20 and 3.28 correlating with C-1 (at δ 67.0 and 68.8) and C-2 (δ 53.9 and 54.7) in HMBC spectrum. Succinic acid has a singlet at δ 2.56 which correlates with the carbonyl group at δ 176.5 in the HMBC spectrum. Multiplets at δ 2.43 and 2.12 were assigned to H-2 and H-3 of proline. Other signals of the pyrrolidine moiety of proline are also observed in the 1H NMR except H-5 which overlapped with other signals in the sugar region. Proline was confirmed by the correlation between H-3 and H-2 of proline with the carbonyl group at δ 174.4 in the HMBC spectrum. Lactic acid shows the quartet of 2 at δ 4.18 and the doublet of H-3 at δ 1.H-36 in 1H and J-resolved NMR spectra. Both signals correlate with a carbonyl
group at δ 179.0 in the HMBC spectrum. The doublet at δ 1.12 which correlate to δ 73.0 in the HMBC spectrum was identified as isopropanol. All of these assignments were confirmed by the NMR spectra of reference compounds.
7.3.3. PCA of beer sample
Among the multivariate techniques, principal component analysis (PCA) is one of the most widely used methods. The concept of PCA is to describe the variance in a set of multivariate data in terms of a set of underlying orthogonal variables (principal components). The original variables (metabolite concentrations) can be expressed as a particular linear combination of the principal components (Eriksson et al., 2001). The principal components can be displayed graphically as a score plot. This plot is useful for observing any groupings in the data set. PCA models are constructed using all the samples in the study. Coefficients by which the original variables must be multiplied to obtain the PC are called loadings. The numerical value of a loading of a given variable on a PC shows how much the variable has in common with that component (Eriksson et al., 2001; Massart et al., 2001). Thus, for NMR data, loading plots can be used to detect the compounds responsible for the separation in the data. Generally, this separation takes place in the first two or three principal components (PC1, PC2 and PC3).
Fig. 7.5A and B show the comparison between the score plots of PCA based on entire range of 1H NMR spectra and projected J-resolved spectra using PC1 versus PC3. Eight components explained 99.9% of the variance, and the first three components explained 93.1% (1H NMR spectra) and 89.3% (projected J-resolved spectra). In the score plot, the similarity of the samples with respect to the intensity of chemical shift correlates with their grouping pattern. The separation of beer samples dramatically increase in the score plot using projected J-resolved spectra. All of the samples are clearly separated.
Fig. 7.5. Score plot of principal component analysis of 2-butanol extract of beer using PC1 vs PC3 based on 1H NMR spectra (A), projected J-resolved NMR spectra (B), and loading plot based on projected J-resolved NMR spectra (C) using entire range of the spectra (δ 0.4 – δ 10.0).
Samples 1–2 have more cholines and less isopropanol, carbohydrates and succinic acid than samples 3–6. Whereas samples 4–5 have more lactic acid, proline and
isopropanol but less succinic acid and cholines than samples 1–3 and sample 6. The non-volatile organic acids, such as succinic acid and lactic acid are both malt-derived by-products of yeast during fermentation. Succinic acid has little effect on the beer flavour except for acidity. On the other hand, lactic acid emerges as the most important non-volatile organic acid in beer since it is the only one present in beer in the above-threshold amounts (Charalambous, 1981; Hardwick, 1994). Proline is always at a high level in beer and influences the colour of final beer through the maillard reaction during wort boiling. The browning reaction is believed to be caused by maltol which is produced by the reaction between proline and maltose (Charalambous, 1981).
Using the PCA analysis based on the entire range of projected J-resolved NMR spectra no significant contribution is observed by the aromatic signals due to their low intensity compared to that of sugars or aliphatic compounds. In order to evaluate their effect on the differentiation, PCA was applied only to the aromatic part of the spectra. Fig. 7.6A shows the score plot thereby obtained using PC2 versus PC3. An eight component model explained 99.9% of the variance while the first three components explained 94.9%. All the beer samples are also well separated. The loading plot (Fig. 7.6B) shows signals of adenine at δ 8.32 and 8.19; uridine at δ 8.00; xanthine at δ 7.90; tyrosine at δ 7.12 and tyrosol at δ 7.02 and 6.70 were found to be the aromatic compounds responsible for the differentiation. Samples 3 and 6 were found to contain more tyrosol but less uridine and tyrosine than the other samples. While samples 1–2 have more uridine, xanthine and tyrosol but less adenine than other samples.
quality of beer. This method will be useful for the product development and quality control.
7.4. CONCLUSION
The fractionation of beer samples by liquid–liquid partition using 2-butanol provided a significant increase in the intensity of aromatic compounds in the NMR spectra. PCA analysis using projected J-resolved spectra for the beer extracts shows increased separation of beer samples compare to conventional 1H NMR spectra. The increased resolution obtained from J-resolved spectra also provided further information on the structures of the compounds in beer.
